Read. Analyze. Sort.
New launch. Clear category.
LAYA reads the chain and sorts the evidence.
One real launch, read and sorted by LAYA
What LAYA read
Four typed questions to the open model
The four categories
No category. A guard veto comes first, so this launch is not sorted.
Sorted by LAYA
Only verified results appear here.
Signed verdict · open weightsInspect result
A real Pons launch from the signed feed. LAYA reads its on-chain facts, runs a fixed guard, asks the open model four typed questions, and sorts the launch into BUY, COPY, SKIP or PASS with the model's real probabilities. A guard veto stops the launch before sorting. Categories are research labels. LAYA never trades.
Built with
What powers LAYA- LAYA open modelOpen weights
Reads the evidence and answers four typed questions.
- NillionChecking
Puts verdicts into words and keeps private notes encrypted.
nilAI status loadingSecretVault status loading - Oasis ROFLChecking
Adds a second, attested signature to each verdict.
ROFL status loading
Words and privacy, never the verdict.Explain the latest verdict with Nillion
Waiting for a verified result from the live feed.
Deterministic checks decide. Partner prose cannot clear a warning.
No partner explanation has been generated for this view.
Source: Nillion nilAI. Model and response time appear only after a real response.
Checking partner services
Save a private signal note
How it works
Every launch, parsed and sorted.
Read the chain. Check the rules, then ask the open model. Sort into one research category.
Read
Start with the chain.
Contract code, holders, curve liquidity, trades, fees and creator history, read at one recorded block. Missing data stays unknown.
Analyze
Rules first, then the model.
A fixed guard checks the launch. The open model answers four typed questions. A veto always wins.
- Actionchoice
- Riskyes or no
- Momentumscore
- Crowdchoice
Sort
One category, full context.
Every probability stays attached. Categories are research labels. LAYA never trades.
- BUYresearch further
- COPYstudy early buyers
- SKIPadverse evidence
- PASSevidence is thin
Decision lab
Walk one real verdict from launch to category.
See what the model chose for each question and what the guard found.
Every option, the model's pick and the GUARD evidence. Finish by verifying it in your browser.
Open the decision labLatest results
Straight off the signed feed.
Newest first, in the order they were made. Nothing picked to look good.
Issue order, guard beside model answer. The lanes plot three hours by GUARD result.
Open weights. Signed verdicts.
Same input. Same weights. Same numbers.
Nothing to take on trust. Rerun any verdict and the numbers match exactly.
Pin the open weights and the exact input. Every probability must match to the last digit.
- Which model decided?
- Open Laya checkpoint, pinned revision and weight hash
- Black box: unnamed, may change quietly
- What did it see?
- Exact input JSON with its SHA-256
- Black box: not published
- Can I rerun it?
- One command, identical probabilities
- Black box: no
- Where are the safety rules?
- A deterministic GUARD, shown above the model
- Black box: mixed into the score
- What gets published?
- Every verdict, including contradictions
- Black box: the wins
- Who trades?
- Only you, from your own wallet, after a confirmation
- Black box: sometimes the bot
Select a signed result to get its replay command.- Status
- Network
- Robinhood Chain / 4663
- Venue
- Pons
- Contract
- Not published
- Desk access
- Free, no token needed
Scoreboard
A thesis until the outcomes say otherwise.
Every verdict is checked again 1, 6 and 24 hours later. Good, bad or waiting, it stays on the board.
The curve is read again at 1, 6 and 24 hours. Pending, unavailable, migrated and failed all stay counted. Mean curve spot return excludes fees and slippage and is not an accuracy claim.